One shared assumption of many recent efforts to delineate a history of fiction (or fictionality, typically understood as a mode of nonliteral reference) is that that term names a conceptual operation, be it intrinsic or culturally learned. This article argues that fiction is merely a particular type of classification, akin but not identical to the classifications performed by terms such as mimesis or verisimilitude. Thus it is nonsensical to claim that fiction qua concept does or does not exist at any given moment: fiction is foremost a way of grouping various literary practices, and it is those practices that emerge over time. The article then recasts the interest in the early novel’s fictionality shown by Catherine Gallagher and others as a problem of practices rather than of concepts. It tracks trends in subject matter and assertions of literal truth through a quantitative diachronic analysis of 230 years of French novels. While these trends cannot by their very nature show the birth of the concept of fiction—which was never born in the first place—they are the type of evidence that should be central to any future history of fiction.